South Equatorial Current of the Indian Ocean

OCEANOLOGICA ACTA- VOL. 17- N°3 ~ -----~1-

Fifty-day oscillation South Equatorial Current Rossbywave Indian Ocean Currents of the Indian Ocean: czcs

Oscillation d'une période de cinquante jours a fifty-day oscillation Onde de Rossby Océan Indien Courants czcs

Karen J. HEYWOOD*, Eric D. BARTON and Graham L. ALLEN

School of Ocean Sciences, University College of North Wales, Menai Bridge, Gwynedd LL59 5EY, UK.

* Present address: School of Environmental Sciences, University of East Anglia, Norwich NR4 7TJ, UK.

Received 24/06/93, in revised form 3/01194, accepted 11102/94.

ABSTRACT Observations of a fifty-day oscillation in the meridional component of the South Equatorial Current in the Western Indian Ocean are presented. Currents measured in 4000 rn of water close to the atoll of Aldabra (46°E, 9°S) reveal the oscillation throughout the water column to a depth of 3 000 m. Further evidence of the oscillation is presented in shipboard Acoustic Doppler Current Profiler (ADCP) sections and in Coastal Zone Col or Scanner (CZCS) imagery. The oscillations are consistent with a zonally propagating Rossby wave caused by shear instability in the zonal flow of. the South Equatorial Current . Oceanologica Acta, 1994.17, 3, 255-261.

RÉSUMÉ Courant sud-équatorial de l'Océan Indien : oscillation d'une période de cinquante jours Une oscillation dont la période est de cinquante jours a été observée dans la composante méridienne du courant sud-équatorial, dans l'ouest de l'Océan Indien. Dans la colonne d'eau de 4000 rn adjacente à l'atoll d'Aldabra (46 °E, 9°S), la courantométrie révèle la présence d'une oscillation jusqu'à 3 000 m de profondeur. Une autre preuve de cette oscillation est apportée par les profils de courantométrie acoustique Doppler (ADCP) et par la télédétection satellitale CZCS. Les oscillations observées sont en accord avec la propagation zonale de l'onde de Rossby produite par l'instabilité du cisaillement dans l'écoulement du courant sud-équatorial. Oceanologica Acta, 1994.17, 3, 255-261.

INTRODUCTION Cutler and Swallow (1984). Current oscillations on shorter · time scales of weeks to months have been noted by Luyten Dominant features of the Indian Ocean include strong and Roemmich (1982) near the African coast and by variability of currents on a range of time scales. The Quadfasel and Swallow (1986) and Schott et al. (1988) off monsoon winds drive seasonal current reversais along the Madagascar. Schott et al. (1988) report fluctuations with a African and Arabian coasts (Luther and O'Brien, 1985) as forty- to fifty-day period in the boundary current transports well as at the equator (Wyrtki, 1973). The South Equatorial off the east coast of Madagascar. An oscillation with a Current variation on the same time scale is documented by period of forty to sixty days bas been reported in the long

255 K.J. HEYWOOD, E.D. SARTON, G.L. ALLEN shore currents of the Somali current off the Kenyan coast (Mysak and Mertz, 1984). Quadfasel and Swallow (1986) present current meter data from a site off the northem tip of Madagascar during March-July 1975 which indicate a fifty-day period oscillation with an amplitude of 12 cm s-1. They also report meanders Bay of a similar amplitude and wavelength about 400 km of in surface current vectors. BengaJ As part of a study of flow disturbance by oceanic islands (Heywood et al., 1990), two current meter moorings were deployed in deep water (4000 rn) west of the coral atoll of Aldabra (Fig. 1). Current records from these moorings, located at 10°S in the South

Equatorial Current, show evidence of a fifty-day Mid lndlan Basin oscillation in the meridional component down to sorne 3 000 rn below surface. Further evidence of current oscillations on a similar scale in the upper layers can be seen in transects of acoustic Doppler current W2 profiler (ADCP) measurements and in Coastal Zone •-9.00 + Co lor Scanner (CZCS) imagery. . W1 ~" -9.25 + Aldabra fs. Such waves may have important consequences for the ~~)\._:;....r ___ .... biological productivity of the area, and more -9.50 particularly for sorne of the isolated islands. We shall Mauritius Basin -9.75 show that waves are clearly defined on a front between 45.00 45.50 46.00 46.50 Longltud• low chlorophyll content water to the north, and greener water to the south, as shown by surface chlorophyll concentration revealed by ocean colour imagery. A Figure 1 regular influx of a greater phytoplankton population may influence the ecology associated with an island, Location of the two moorings, Wl and W2, around the coral atoll of Aldabra. Depth contour of 1 000 m is shown as a dashed Une in the inset. such as Aldabra, by enhancing the growth of zoo plankton and higher organisms. the loss of sorne data sets. The data were edited to eliminate spikes, rotated by 6° to correct for the magnetic CURRENT METER OBSERVATIONS deviation from true north, and low-pass filtered using a Cosine-Lanczos 131-point filter with a half-power point at 40 hours in order to eliminate tidal and other high Two current meter moorings, Wl and W2, were deployed frequency variability. During filtering, the data were in 4000 rn of water to the northwest of Aldabra (46° 20' E, decimated to six-hourly values. As attention is focused on 9° 25' S; Fig. 1) from RRS Charles Darwin in late April low frequency phenomena, the data were further filtered by 1987 and retrieved early in July 1987, approximately applying a 21-point running-mean filter, removing eleven weeks later. Six Aanderaa current meters were oscillations with period less than five days. deployed at each mooring, although battery failure led to The mooring positions, post-filter data record lengths, means and standard deviations of the current components Table 1 are shown in Table 1. The prevailing neac-surface current Statistics ofthefiltered time series from current meter moorings Wl and W2. is westward (W2) or northwestward (Wl) as one would expect for the South Equatorial Current during the austral autumn and winter. The standard deviations of the Mooring Depth Length Zonal Component Meridional Component meridional components are considerably larger than the (rn) of data mean std dev. mean std dev. (days) (cm s"1) (cm s"1) (cm s"1) (cm s"1) me ans. This implies the possibility of a significant oscillation in this component.

W1 60 69 -3.9 6.8 3.9 5.6 At W1 a clear fifty-day oscillation, primarily in the 9° 16' s llO 69 -3.9 8.1 0.8 6.0 meridional component, is seen at 500 rn (Fig. 2). What 45° 58' E 300 39 5.3 5.2 -3.6 6.9 appears to be part of a similar oscillation is evident at 300 500 69 3.0 5.8 -3.3 6.6 rn; the signal is in phase between the two levels and has the 3000 8 -6.4 1.0 -5.3 1.3 same amplitude, around 10 cm s-1. The low-frequency W2 80 69 -17.0 10.9 -0.6 11.4 oscillation is barely evident at the two uppermost Ieve1s. At 9°0' s 130 69 -12.5 12.5 - 1.0 10.5 W2, an oscillation of the same period is seen at ali levels 45° 27' E 330 69 - 1.5 13.9 • 3.9 12.5 (Fig. 3), although higher frequency fluctuations also 480 69 2.2 9.6 • 6.0 10.3 1000 69 2.5 4.7 -4.2 6.5 influence the upper two levels. The oscillation is evident 3000 69 -0.8 2.9 1.5 3.4 even at the deepest observed level with the same frequency

256 FIFTY-DAY OSCILLATION

'., 20 1 60 E 0 ~itJiiiil.~ll\lllia-.illlllll~~__..~ u -20

110m 130 m '., 201 E 0 u -20

300 m "' 20 l '., 20 j E 0 u ~ ol -20 -20 460 m 500 m ' 20 j "' 0 ·., ~ E u :1 -20 -20 1000 m ' 20 j "' 0 110 120 130 140 150 160 170 180 190 ~ Time (days) -20 Figure 2 2 ·- 3000 m Carrent ve ctor ti me series at mooring W 1 at the depths shown. oj ' Northward currents point up the page. These data have been smoothed e-- o ....._...~ ~•gp.. r~r~wwn~ ... -- ~uc.œ.. n-..------=- .. and filtered as described in the text. " -20 and period. The amplitude of the oscillation decreases 110 120 130 140 150 160 170 180 190 from about 20 cm s· l above 330 rn to less than 10 cm s· 1 at Time (days) 3 000 m. The phase at the deepest level, however, is 180° with respect to ail other levels and the mean flow direction Figure 3 is also reversed (Tab. 1). Although we initially suspected instrumental error, we do not believe this to be the case; a Carrent vector lime series at mooring W2. current meler record also at 3 000 rn at mooring Wl ended prematurely after only two weeks, but is sufficiently long moorings separately, weighting each record equally. The to indicate the same 180° phase reversai between the EOFs of the zonal components showed no evidence of the currents above 1 000 rn and tho se at 3 000 m. This suggests oscillation and so are not discussed further. At W 1, the a baroclinic component to the wave. Note that the presence two shorter records are ornitted from the EOF calculations, of the oscillation at this great depth is unprecedented. so the components at W 1 are decomposed into three The temperature time series (not shown) of the two upper orthogonal modes. The EOFs calculated for the current current meters at both moorings exhibit an oscillation with components at W2 are each based on the records from all a period of the order of 25 days. The amplitude of the six depths, yielding six modes. Almost ali of the variance oscillation is about 4°C and is consistent with the is explained by modes 1 and 2. explanation that the current meters are pulled down by the At mooring W2 the meridional component is dorninated by lateral straining due to the currents. Pressure sensors on the first mode, which accounts for almost 90 % of the severa! meters indicated that vertical di splacements were total variance (Tab. 2) and clearly shows the fifty-day usually Jess than 10 rn but on two occasions were as large oscillation (Fig. 4). The vertical structure of mode 1 (Fig. 5) as 20 rn during periods of stronger currents. Although this shows only a sli ght reduction in the magnitude of the vertical motion does affect the temperature signais, it has eigenfunctions with depth. Mode 2 appears baroclinic with negligible effect on the oscillation observed in the current a zero crossing at 250 rn , has a periodicity of around measurements themselves si nce there is relatively little 25 days but is not significant because it contributes only shear. 7 % of the total variance. Fewer data at W 1 produce Jess clear results than at W2, but common characteristics are still evident. The first EMPJRJCAL ORTHOGONAL FUNCTIONS mode, which accounts for only 50 % of the variance, reveals the fifty-day oscillation but hi gher frequency Empirical Orthogonal Functions (Kundu et aL, 1975) were fluctuations modify the low frequency signal (Fig. 4). The calculated for each of the current components at both second mode is similar to that at W2 but is more

257 K.J. HEYWOOD, E.D. BARTON, G.L. ALLEN

Table2 -1 0 -1 0 0 0 Empirical orthogonaljunctions at carrent meter moorings Wl and W2...... ·----... , ·-- ,, ...:] ...... 500 !lOO MooringWl Model Mode2

Explained variance 50 37 1000 1000 (%) :§: :§: 1500 t 1500 Eigenfunction 60m 0.33 -0.67 t llO rn 0.68 -0.32 0" 0" 500m 0.65 0.67 2000 2000

MooringW2 Model Mode2 2500 2500 Explained variance 90 7 (a) (b) Eigenfunction SOm 0.47 -0.75 3000 3000 l30m 0.45 -025 330m 0.54 0.41 480m 0.44 0.43 Figure 5 lOOOm 0.26 0.13 Eigenjunctions of the meridional components at (left) mooring Wl and 3000m -0.14 -0.12 (right) mooring W2. EOFsfor mode 1 are represented by trianglesjoined by a fallline and for mode 2 by squares joined by a dash line.

40 Mooring 11'2 Mode 1

20 As the EOFs for W1 are based on only one time series from depth and two from the upper 150 rn, it is not e"' 04--.-.--~.-.--.-.-,r+.-.--r-.-.~~~· surprising that the fifty-day signal in the modal amplitudes (.) at Wl is less apparent than at W2. The EOF results reflect -20 the emphasis on the upper layers in the data rather than an absence of the fifty-day oscillation. -40 ~!i 200 l--M·o-o~r-in~g~W'2-,M~o~dre-2.-~~~--~--~-n-r-. SPATIAL CURRENT MEASUREMENTS " ' \;-,()'="'"' ~~ During the two-three day passage legs to/from Aldabra

-20 (Fig. 6), continuous near-surface current measurement was undertaken using the shipboard 150 kHz ADCP. Absolute 200 currents were calculated using satellite navigation and the ~!i l--M·o-o•r+in•gr-Wr1~M-o~d~e~1-.--r7~hr-r-k-,,-.-. ship's electromagnetic log, allowing for misalignment of the ADCP relative to the ship's heading (Pollard and Read, "·~~ 1989). During the passage legs a constant speed of about -20 10 knots (approx. 5 rn s-1) was maintained and no tums Mooring 11'1 Mode 2 were made, so gyro compass swings should be negligible. The ADCP data were despiked and interpolated to give values every 5 km, and then smoothed by applying a 21- point running mean. Currents were then depth averaged over the top lOO rn to suppress small scale fluctuations and the overall mean on each leg was subtracted. Currents throughout the depth range of the ADCP (near surface to 110 120 130 140 150 160 170 180 190 400 rn) showed similar structure. lime (days) Section A was undertaken from Mauritius to Aldabra, 13- Figure4 17 April, and is about 1 000 km long. The current vectors Expansion coefficient series of modes 1 and 2 of the EOFs for meridional show a wave of length between 400 and 600 km, and carrent components at Wl and W2, anits are arbitrary. magnitude greater than 20 cm s-1. The smaller scale variations also evident could be the signatures of mesoscale eddies (Rossby radius approximately 70 km). significant at Wl where it contributes 37 % of the variance Section B, along the track from Aldabra to the Amirante (Tab. 2). The vertical structure (Fig. 5) of both modes is Trench between 3-6 May, shows a distinct oscillation in consistent with the analysis at W2. The shallowest level the currents of amplitude 15 cm s-1 and apparent shows reduced amplitude because of the effect of higher wavelength about 400 km. These wave amplitudes seen on frequency variability. sections A and B are in agreement with the magnitude of

258 FIFTY-DAY OSCILLATION

DISCUSSION

By correlating the meridional current components at Wl 10 500 rn and W2 480 rn, where the fifty-day oscillation is clear at both sites, an estimate of its speed of propagation may be made. The correlation coefficient is 0.8 (significant at 95 % confidence level) with a lag of 150 hours; Wl leads W2, i.e. westward propagation. Taking the distance over which the disturbance travels in this time to be 53 km (the zonal separation between W1 and W2) yields a propagation speed of 0.1 rn s-1 westwards. We believe this to be the first direct measurement of the propagation speed of the fifty-day oscillation. The characteristics of the oscillation identified in the current time series are consistent with the presence of a zonally- propagating Rossby wave. For such a wave, the zonal wave number, K, is related to the frequency, w, by the dispersion equation:

45 60 Longitude (E) -- 20 cm s-1 Figure 6 where f3 is the latitudinal gradient of the Coriolis Passage legs A and B during which acoustic Doppler current profiler parameter, f, and g is gravitational acceleration. The estimates of near-surface currents were obtained. Section A- Mauritius to barotropic (depth independent) mode is that with n 0 and Aldabra, 13-17 April, 1000 km long. Section B: Aldabra to Amirante = Trench, 3-6 May, 600 km long. Depth averaged ADCP currents in the ho is theo the water depth. The equivalent water depth upper 100 m. The mean currents for each leg have been subtracted. is given by hn = Ap Hlp for baroclinic (internai) modes n = 1, 2, etc. and layer depth H. We note that the phase speed is given by c = w/K. the fifty-day oscillations of about 20 cm s- 1 observed at moorings Wl and W2 in the upper few hundred metres. If we take the frrst baroclinic mode, i. e. n = 1 and assume one layer 500 rn thick with density difference 6 kg m-3, and the phase speed of the wave, c = 0.1 rn s- 1 as calculated above, theo the wavelength (2 rc/K) for a SATELLITE IMAGER Y zonally- propagating Rossby wave at 9°S would be 425 km, and the associated period 49 days. If the wave were Although during the period of the experiment, no clear barotropic, taking ho = 4 000 rn, theo the resulting satellite imagery was available, a number of CZCS images wavelength and period are not significantly different from in other years bas been obtained. Sorne of these, ali from the first mode baroclinic case. The fact that the meters at the second quarter of the year, show oscillations in 3 000 rn are 180° out of phase with the near 'surface the South Equatorial Current. A scene from 5 May 1986 oscillation is evidence of the baroclinic nature of these (Fig. 7 a) reveals a wave like structure in surface pigment oscillations. The waves might be generated by barotropic extending northwestwards from the northern tip of instability of currents trapped close to the surface, but Madagascar and passing close to Aldabra. The higher become baroclinic on propagating away from the area of chlorophyll content waters seem to be associated with the generation. current passing close to Madagascar, and are theo carried Priee and Rossby ( 1982) discuss observations of such downstream beside more oligotrophic waters to the north, planetary waves in SOFAR float data at 1300 rn in the allowing the wave to be seen on the front between the two. western North Atlantic. Wavelength (340 km) and period A similar structure is seen in the corresponding image for · (61 days) are consistent with a barotropic wave, but they 30 March 1979 (Fig. 7 b). The wavelength appears to be include the effect of changes in topography. They note about 250 km, somewhat shorter than indicated by the . however that the gravest baroclinic mode has similar other observations. The reason for this is not clear, and characteristics; they conclude that stratification does not there are no image sequences to allow investigation of strongly influence the wave. They suggest that these waves propagation or development of the feature. The final are caused by baroclinic eddies in the Gulf Stream image shown [26 April 1979 (Fig. 7 c)] shows a more evolving towards barotropy. cusp-like feature whose wavelength is larger, about 400 km. In ali three images the wave extends in the mean Kindle and Thompson (1989) calculate baroclinic Rossby direction of the South Equatorial Current at this time of waves in the southwest Indian Ocean to have wavelength year. These images provide strong evidence for the 800 km and period 52 days (they assume a different existence of wave like perturbations near Aldabra. stratification to our results). They suggest that the 400 km

259 K.J. HEYWOOD, E.D. BARTON, G.L. ALLEN

Longitude (E) degrees Figure 7 4 5' 46' 47' 48' 49' Coastal Zone Color Scanner images showing chlorophyll content of the surface waters. Th e dark blue signifies cloud or land; black denotes th e lowest phy toplankton concentration. a) 5 May / 986; b) 30 March 1979; c) 26 April 1979. The colour scale is given with (c) .

waves o bserved in the surface c urre nts by Quadfasel and Swallow (1986) are not associated with the barotropi c instability causing the fifty day oscillation in their current meter data (and in Kindle and Thompson's one layer model) , but are a mani festati on of a baroclinic instability possible in westward flow away fro m the equator (> 10°). Although our observations of surface current oscillati ons are consiste nt w ith th ose o f Qu adfasel and Swallow (1986) and are closer to the equator, it is impossible to know definiti vely the generati on mechani sm from thi s data set and Longitude (E) degrees further modelling and observations are warranted. 44' 4 5' 46' 47' 48' The estimate of the wavelength deri ved from the ADCP secti ons is unaffected by the ship's speed, 5 rn s- 1, which is an order of magnitude greater than the phase speed of the wave, 0.1 rn s- 1. The ADCP sampling of the wave may -therefore be considered synoptic, i .e. any change in the apparent wavelength due to the Doppler effect is .."' small compared to the true wavelength . If the ~ .."' meridi onal osci.ll ati ons are assumed to be Rossby '0 waves , one would expect them to be propagating ~ .. westwards. During ADCP section A, the ship's '0 :> i 1 heading was approximately 3 15°, so the apparent -' 600 km along-track wavele ng th could be corrected to a zonal wavelength of 400 km. Section B lies east-west so no correction would be necessaty to the observed wavelength of 400km. Although there have been earlier reports o f similar low frequency current oscillations in the Indian Ocean , their origin is not fully understood.

Chlorophyll (mg/ m 3 ) Production of these forty- to sixty-day current oscillations was originally attributed to the Madden and Julian oscillation (Madden and Julian, 1972), a forty- to sixty-day oscillation in the tropical wind field (Mysak and Mertz, 1984). Longitude (E ) degrees However, modelling results (Woodberry et al. , 43' 44' 45' 47' 1989; Kindle and Thompson, 1989) indicate that direct wind fo rcing may not be the d riving mechanism. ln the numeri cal models the area to the west of 50°E and between the Equator and .. 10°S is identified as one of intense eddy .. generation. lnstabilities in the horizontal shear ..5 'C give rise to forty to sixty day fluctuations in the ~.. currents . The mode] described by Schott et al . '0 i! ( 1988) shows such instabilities occurring just to ~ the north and northwest of Madagascar. Severa] times each year thi s extended out to 55°E and to 11 os. Our observations are further north and further west.

260 FIFTY-DAY OSCILLATION

It appears that there are preferential times of year when 3 000 rn, and have enabled us to measure the westward these waves are generated. The reduced gravity model of propagation speed for the first time. The characteristics Kindle and Thompson ( 1989) showed that Rossby waves of the oscillation as seen in the current meter records and were generated by a barotropic instability associated with shipboard ADCP may be explained by a zonally the east African Coastal Current, beginning around April propagating Rossby wave. each year (since this is a one layer model, baroclinic As was seen in the CZCS images, these waves can have instabilities are not possible). The model of Schott et al. important ramifications for the phytoplankton distributions (1988) also showed the waves to be sporadic. Our in the region. Ocean colour imagery is more useful than observations happen to fall in the austral autumn and A VHRR or ATSR imagery in observing the wave because winter periods when waves are generated. Although we do of the uniform sea surface temperatures in the region. The not have current meter measurements from other times of future availability of improved ocean colour observations year, we do have CZCS images from other seasons of the from the SeaWiFS satellite will allow further study. year; the images showing waves are from March, April and May. Large meanders have been seen in drifting buoy trajectories in this area (Molinari et al., 1990) but there are too few to show the organised oscillation we have Acknowledgements discussed here or to indicate any seasonal dependence. Earlier observations of the fifty-day oscillation have been This work was supported by UK Natural Environment confined to single sites in coastal regions and to the Research Council grants GR3/5549 and GR3!7368. We are upper layers. Our observations show that it also occurs grateful to Jeff Whiting of the NASA Goddard Space away from the ocean boundaries and at depths as great as Flight Center for providing the CZCS images.

REFERENCES

Cutler A.N. and J.C. Swallow (1984) Surface currents of the lndian of surface buoy trajectories. J. geophys. Res., 95,7217-7238. Ocean. lnstitute of Oceanographie Sciences Report No. 187, 8 pp. Mysak L.A. and G.J. Mertz (1984) A 40- to 60-day oscillation in Heywood K.J., E.D. Barton and J.H. Simpson (1990) The effects the source region of the Somali current during 1976. J. geophys. Res., of flow disturbance by an oceanic island. J. mar. Res., 48, 55-73. 89,711-715, 1984. Kindle J.C. and J.D. Thompson (1989) The 26- and 50-day Pollard R.T. and J.F. Read (1989) A method for calibrating ship oscillations in the Western Indian Ocean: mode! results. J. geophys. mounted acoustic Doppler current profilers, and the limitations of Res., 94,4721-4736. gyro compasses. J. atmos. ocean. Technol., 6, 865-959. Kundu P., J.S. Allen and R.L. Smith (1975) Modal decomposition Priee J.F. and H.T. Rossby (1982) Observations of a barotrpic of the velocity field near the Oregon coast. J. geophys. Res., 5, 683- planetary wave in the western North Atlantic. J. mar. Res., Suppl., 704. 40, 543 - 558. Luther M.E. and J.J. O'Brien (1985) Modelling the variability in Quadfasel D.R. and J.C. Swallow (1986) Evidence for 50-day the Somali Current, in: Mesoscale!Synoptic Coherent Structures in period planetary waves in the South Equatorial Current of the Indian Geophysical Turbulence, J.C.J. Nihoul and B.M. Jamart, editors. Ocean. Deep-Sea Res., 33, 1307-1312. Elsevier Science Publications, Amsterdam, The Netherlands. Schott F., M. Fieux, J. Kindle, J, Swallow and R. Zantopp (1988) Luyten J.R. and D.H. Roemrnich (1982) Equatorial currents at the The boundary currents east and north of Madagascar. 2: Direct semi-annual period in the lndian Ocean. J. phys. Oceanogr., 12, 406- measurements and model comparisons. J. geophys. Res., 93, 4963- 413. 4974. Madden R.A. and P.R. Julian (1972) Description of g1obal-scale Woodberry K.E., M.E. Luther and J.J. O'Brien (1989) The wind circuhition cells in the tropics with a 40-50 day period. J. atmos. Sei., driven seasonal circulation in the Southern Tropical Indian Ocean. J. 29, 1109-1123. geophys. Res., 94, 17985-18002. · Molinari R.D., D. Oison and G. Reverdin (1990) Surface current Wyrtki K. (1973) An equatorial jet in the Indian Ocean. Science, distributions in the tropical Indian Ocean derived from compilations 181, 262-264.

261